|
Journal of Cancer Research and Therapeutics, Vol. 6, No. 3, July-September, 2010, pp. 267-271 Original Article Mapping the extent of disease by multislice computed tomography, magnetic resonance imaging and sentinel node evaluation in stage I and II cervical carcinoma S Rajaram1, H Sharma1, SK Bhargava2, RP Tripathi3, N Goel1, S Mehta1 1 Department of Obstetrics & Gynaecology, University College of Medical Sciences & Guru Teg Bahadur Hospital (UCMS & GTBH), Delhi, India Correspondence Address:S Rajaram, Department of Obstetrics & Gynaecology, University College of Medical Sciences & Guru Teg Bahadur Hospital, Delhi-110095, India, rajaram.shalini@gmail.com Code Number: cr10061 PMID: 21119251 DOI: 10.4103/0973-1482.73342 Abstract Aims: (1) To map the extent of disease in women with stage I and II carcinoma cervix by multislice spiral computed tomography (CT), magnetic resonance imaging (MRI) and sentinel nodes. (2) To assess accuracy of each modality individually and in conjunction with FIGO clinical staging.Design and Setting: Prospective, single-blind study. Departments of Obstetrics and Gynaecology, Radiodiagnosis, and Pathology, UCMS and GTBH and Division of Radiological Imaging and Bioinformatics, INMAS, Delhi. Material and Method: The study was conducted on 25 women with cervical cancer FIGO stage I and II. Each woman underwent clinical staging, multislice spiral CT and MRI which was compared to the gold-standard histopathology/cytology. The overall accuracy of each modality and improvement of clinical staging by CT/MRI were noted. Sentinel nodes were evaluated by intracervical Patent Blue V dye injection. Statistical Analysis: Sensitivity, specificity, positive and negative predictive values were calculated by 2Χ2 contingency tables. Results: The accuracy of staging by FIGO, CT and MRI was 68%, 52% and 80%, respectively. MRI and CT improved the overall accuracy of FIGO staging to 96% and 80%, respectively. Sentinel nodes were identified in 89% of patients with 91% accuracy. Conclusion: MRI emerges as the most valuable stand-alone modality improving accuracy of FIGO staging to 96%. Sentinel lymph-node evaluation appears promising in evaluating spread beyond cervix. Keywords: Carcinoma cervix, CT, MRI, sentinel node, FIGO staging Introduction Cancer cervix is one of the leading causes of cancer death among women in developing countries. Discrepancies of approximately 25% in early stages and 65-90% in advanced stages have been noted on comparison of clinical with surgical staging. [1] The greatest difficulties in clinical evaluation are in the assessment of invasion of the parametrium and the pelvic side walls, in the estimation of tumor size (especially in endocervical growth) and in the evaluation of lymph-node metastasis. [2] Clinical evaluation provides little information about extension to vesicocervical and rectovaginal spaces, both of which must be free for a radical hysterectomy to be completed successfully. [3] To improve accuracy of clinical staging, various imaging modalities such as computed tomography (CT) and magnetic resonance imaging (MRI) have been used but findings have not been translated into clinical practice. CT and MRI can overcome the deficiencies in FIGO staging by precisely documenting tumor size, invasion, parametrial extension, lymph-node involvement and bladder or rectal involvement, thereby altering the therapeutic protocols. [4] Besides, MRI has been shown to be a cost-effective preoperative staging technique requiring fewer tests and procedures compared with those who undergo standard clinical staging. [5] Sentinel lymph node (SLN) evaluation is an established technique for melanoma and breast carcinoma and its use is recent in vulval, lung and cervical carcinoma. [6],[7],[8],[9] The incidence of positive nodes in the most favorable group of patients, stage I B, is approximately 15%. [10] This means that more than 80% of these patients who undergo lymphadenectomy derive no benefit from the procedure, yet must endure the associated increase in operative time, blood loss, risk of lymphocysts and lymphedema. Thus SLN evaluation is a practical approach to reduce extensive lymph-node dissection. [2] The present study was done to map the extent of disease in stage I and II carcinoma cervix patients by multislice spiral CT and MRI and to assess accuracy of each modality individually and in conjunction with FIGO staging. In the absence of parametrial infiltration by CT/MRI women with clinical stage II B disease were offered surgical therapy. This is of value in developing countries like India where a long waiting period for definitive radiotherapy exists. [11],[12] Lymph-node mapping by sentinel node evaluation when used in conjunction with radical hysterectomy may offer a less aggressive approach and decrease surgical morbidity. Materials And Methods The study was carried out on 25 women of carcinoma cervix stage I and II in the Departments of Gynecology, Radiodiagnosis and Pathology, UCMS and GTBH, and Division of Radiological Imaging and Bioinformatics, INMAS, Delhi, India. All women underwent multislice spiral CT on Somatom Zoom plus 4 machine, Siemens, Munich, Germany. Both unenhanced and post contrast films were taken with 5-mm slice thickness and thinner slices whenever required. Data acquisition was divided into two parts Arterial phase (40 seconds after injection of contrast material) and venous phase (180 seconds after injection of contrast material). Scanning was performed during the breath-hold in deep inspiratory phase. The MRI was performed on a high-field 1.5-T magnet system (Magnetom), Siemens, Munich, Germany. High-resolution sagittal T1-weighted spin-echo images and axial and sagittal T2-weighted fast spin-echo images were obtained by using a phased-array coil with 5-mm slice thickness. Thinner sections and contrast studies were done when required. Criteria used for interpretation of the results by both imaging modalities were: a) Cervical tumor appears as a low attenuation mass on CT while on T2-weighted MR images, it is seen as a intermediate signal intensity mass disrupting low signal intensity fibrous cervical stroma.b) Presence of all pelvic lymph nodes along the iliac (common, internal and external) chains was noted. A maximal axial diameter of >10 mm and/or the presence of central necrosis was used as the criteria for diagnosis of metastasis by both CT and MRI. The central necrosis was defined as central density of <20 horsepower unit. c) Criteria used for para-aortic lymph-node metastasis by both CT and MRI was short-axis diameter >1cm. d) Vaginal invasion was seen as wall thickening or eccentric mass on CT while on MRI it appears as loss of low-signal intensity in the vaginal walls. e) Parametrial invasion was seen as ill-defined lateral cervical margin, parametrial stranding, periureteral fat obliteration or any eccentric parametrial mass on CT imaging and as abnormal signal intensity in parametrial region with obliteration of low-signal intensity of cervical stroma on MRI. f) The presence of fat planes between tumor and the bladder or rectal wall was used as exclusion criteria for involvement of these structures. CT and MRI in all cases were done by the same radiologist in their respective departments and were blinded to the clinical stage. Likewise radiologist performing CT examination was blinded to MR findings and vice versa. The surgery was performed within 2 weeks of imaging. Sentinel node evaluation was done before radical hysterectomy. Women were placed in low lithotomy position after anesthesia and abdomen was opened by midline vertical incision. The peritoneum was opened to expose the retroperitoneal vessels and accompanying lymphovascular channels taking care not to disrupt the lymphatics. At this point, using 25-gauge spinal needle, 1-2 ml of 2.5% solution of Patent Blue V dye was injected into the four quadrants of the cervix around the tumor. Blue lymphatics emerging from the lateral parametrium were followed till they ended in a blue node, which was considered as the SLN. Routine lymphadenectomy was then completed. The number and location of sentinel node was recorded on both sides and sent separately for hematoxylin and eosin (HandE) stains. Negative nodes were evaluated by immunohistochemical staining for epithelial membrane antigen to rule out micrometastasis. Clinical stage II B patients with parametrial involvement either on CT or MRI were not operated and underwent TVS-guided FNAC from bilateral parametrium. Histopatology/cytology constituted the gold standard for statistical analysis. All clinical stage I, II A and II B with no parametrial involvement on CT/MRI underwent type III radical hysterectomy with sentinel node evaluation. Clinical stage II B patients with parametrial involvement on CT/MRI received neoadjuvant chemotherapy followed by radiotherapy. Statistical analysis Sensitivity, specificity, positive and negative predictive values were calculated by 2Χ2 contingency tables. Overall staging accuracy, understaging and overstaging was calculated. Results Clinical staging Tumor was not seen in two patients on clinical examination (in one patient lesion was too small and the other had an endocervical growth). Parametrial assessment was accurate in 37 of 50 parametrial regions (25 right and 25 left). The overall staging accuracy by clinical examination was 68% (17 patients correctly staged). Three patients (12%) were understaged and five patients (20%) were overstaged by FIGO clinical staging. CT and MRI staging CT failed to detect tumor in five patients because the lesions were either small exophytic, ulcerative or endocervical. One patient had endocervical growth detected only on MRI [Figure - 1]. Parametrial assessment by CT was accurate in 41 of 50 regions with accuracy of 82%. CT accurately staged 13 of 25 patients (overall staging accuracy 52%), while 20% and 28% were understaged and overstaged, respectively. MRI detected tumor in all 25 patients (100% accuracy). Parametrial assessment was accurate in 43 of 50 regions by MRI (86% accuracy) [Figure - 2]. The overall accuracy of MRI staging was 80% (20 of 25 patients), 4% were understaged and 16% were overstaged. Clinical staging (FIGO) + CT/clinical staging (FIGO) + MRI: The overall accuracy of staging by FIGO was 68% (17 of 25 patients). Of the five patients overstaged by FIGO staging, CT correctly staged two while MRI correctly staged four by accurately assessing the parametrium. Three patients understaged by FIGO staging were all correctly staged by MRI, while one patient was correctly staged by CT. Thus, it was observed that MRI complements FIGO staging by improving the overall accuracy of FIGO to 96% (seven out of eight patients were correctly staged by MRI) and that CT improved FIGO staging to 80% (three of eight patients were correctly staged by CT) [Figure - 3]. Vertical extension into the uterus, vesicocervical and rectovaginal extension: Of the nine operated patients, vertical extension was documented in three patients histopathologically, out of which CT/MRI assessed extension in one patient with accuracy of 77.7%. In one of the operated patients of FIGO stage I B2, hysterectomy was abandoned because of involvement of both vesicocervical and rectocervical spaces not documented on clinical examination. MRI correctly assessed both spaces with accuracy of 88.89% in vesicocervical spaces and 66.67% for rectovaginal space assessment. CT documented only vesicocervical space involvement in that patient but not the rectocervical space obliteration with accuracy of 100% for vesicocervical and 66.67% for rectovaginal space assessment. Sentinel node evaluation: A total of 11 SLNs were removed in nine operated patients (identification rate 89%). The most common location for sentinel node was obturator basin. The accuracy of sentinel node detection was 91%. The sensitivity, specificity, positive and negative predictive values were 66.6%, 100%, 100% and 88.8%, respectively. Side effects of patent blue V: No patient had hypersensitivity reaction to the dye. All patients had blue green discoloration of the urine for 24-48 hrs. In one patient, there was systemic absorption of the dye, which led to the false lowering of pulse oximetry but reassuring oxygen saturation was obtained by arterial blood sampling. All lymph nodes negative for metastasis on HandE staining were also negative on immunohistochemical staining for epithelial membrane antigen. Discussion The FIGO staging system is used universally for staging carcinoma cervix but discrepancies exists between clinical and surgical staging. [1] In search for more accurate diagnostic tools, several studies have been published using CT or MRI but have not been routinely applied in clinical practice. When clinical staging, CT and MRI were used as stand-alone modalities, the staging accuracy was 68%, 52% and 80%, respectively in the present study. However, when imaging modalities were used in conjunction with clinical staging, MRI was clearly ahead increasing accuracy of FIGO staging from 62 to 96%, while CT improved it to 80%. To the best of our knowledge, the improved FIGO staging when combined with MRI/CT has not been studied to date, though individual staging accuracy has been described previously. In the study by Kim et al. on 30 patients of carcinoma cervix, the overall staging accuracy by FIGO, CT and MRI were 70%, 63% and 83%, respectively, comparable to the results in the present study. [13] Ozsarlak studied correlation of preoperative CT, MRI and FIGO staging with histopathology findings in cervical carcinoma in 36 patients. They documented the overall accuracy of staging for clinical examination, CT and MRI to be 47%, 53% and 86%, respectively. [14] Tumor detection was 100% with MRI in the present study and was found to be better than CT (80%) and clinical staging (92%). This is the highest accuracy of MRI for tumor detection reported so far. CT can miss small tumors which are isodense with normal cervical tissue but this can be improved by spiral and multislice CT techniques as was done in the present study. Hricak et al. documented tumor detection of 91% with MRI, while Kim et al. also stated that tumors were more consistently identified by MRI than CT (80% vs 67%). [4],[13],[15] Parametrial evaluation by CT is 30-50%; normal parametrial vessels and ligaments can be misidentified as parametrial extension. This can be improved by contrast enhanced and thinner sections as in the present study. [16] Kim et al. documented the accuracy rates of 78%, 70% and 92% for clinical staging, CT and MRI, respectively, for parametrial evaluation in 20 patients of carcinoma cervix. [13] Hricak et al. in their review article have highlighted the fact that the strength of MRI is its high negative predictive value making it useful in selecting candidates for surgery. [2] Similarly, in the present study, MRI had a high negative predictive value (97%) for parametrial assessment. Ozsarlak et al. concluded MRI to be more accurate than CT in evaluating the locoregional extension in cancer cervix. [14] As both imaging modalities rely on the same morphological criteria for lymph-node involvement, the detection rate (88.9%) was same by both CT/MRI in the present study. Ozsarlak documented accuracy rate of 75% for MRI and 86% for CT in lymph-node evaluation. [14] Vertical extension of tumor into the uterus although disregarded by FIGO for staging, lowers 5-year survival of stage I B and II A patients by 10-20% with a two-fold greater incidence of distant metastases. [17] In the present study vertical extension proven by histopathology was seen in three of nine operated patients and was documented in one patient by both CT and MRI. Vesicocervical and rectovaginal spaces are clinically difficult to evaluate but have important surgical implications. In one patient hysterectomy had to be abandoned because of obliteration of vesicocervical and rectovaginal spaces. The rectovaginal space involvement was identified only by MRI. Thus MRI emerges as an important stand-alone modality reliably assessing various parameters namely tumor, tumor size, parametrial extension, lymph nodes, vertical extension and vesicocervical and rectovaginal space involvement. In addition it provides a cost-effective approach eliminating need for cystoscopy or sigmoidoscopy minimizing cost per patient. [5] A total of 11 sentinel nodes were removed with mean number of sentinel nodes of 1.2 per patient. Similar identification rate has been seen in most studies with the mean number of sentinel nodes removed being 1.8 and 1.7. [18],[19] The identification rate was 89% with patent blue V dye in the present study. Various studies in the past have documented the sentinel node yield ranging between 60 and 100% regardless of the surgical route or the detection method. [19],[20] Dargent et al. documented 100% identification rate using blue dye alone for laparoscopic sentinel node detection. [18] Recent studies suggest the addition of radioactive colloid to blue dye to increase the identification rate without added side effects. Verheijen et al. documented SLN detection rate of 80%, while Zhang et al. showed 100% identification rate by radioactive colloid. [21],[22] The present study using blue dye alone showed feasibility for sentinel node identification. However, more studies using adequate blue dye (4 ml) are warranted. [18],[20] References
Copyright 2010 - Journal of Cancer Research and Therapeutics The following images related to this document are available:Photo images[cr10061f1.jpg] [cr10061f2.jpg] [cr10061f3.jpg] |
|